Identification of Pathogenicity-Related Effector Proteins and the Role of Piwsc1 in the Virulence of Penicillium italicum on Citrus Fruits

Blue mold caused by Penicillium italicum is one of the two major postharvest diseases of citrus fruits. The interactions of pathogens with their hosts are complicated, and virulence factors that mediate pathogenicity have not yet been identified. In present study, a prediction pipeline approach based on bioinformatics and transcriptomic data is designed to determine the effector proteins of P. italicum. Three hundred and seventy-five secreted proteins of P. italicum were identified, many of which (29.07%) were enzymes for carbohydrate utilization. Twenty-nine candidates were further analyzed and the expression patterns of 12 randomly selected candidate effector genes were monitored during the early stages of growth on PDA and infection of Navel oranges for validation. Functional analysis of a cell wall integrity-related gene Piwsc1, a core candidate, was performed by gene knockout. The deletion of Piwsc1 resulted in reduced virulence on citrus fruits, as presented by an approximate 57% reduction in the diameter of lesions. In addition, the mycelial growth rate, spore germination rate, and sporulation of ΔPiwsc1 decreased. The findings provide us with new insights to understand the pathogenesis of P. italicum and develop an effective and sustainable control method for blue mold.


Introduction
Citrus fruits are one of the most important commercial fruits with an annual yield over 124.3 million tons worldwide [1]. Postharvest citrus fruits are extremely susceptible to fungal infection, particularly during storage and transportation [2]. In China, citrus blue mold caused by Penicillium italicum results in about a 30-50% loss of annual citrus production [3]. The use of chemical fungicides, such as prochloraz and imazalil, has been the primary approach to control blue mold in citrus fruits [4]. However, extensive or intensive applications of these chemicals leads to the emergence of resistant fungal populations and poses risks to human health and the environment [5]. Although biological control has been proposed as an alternative strategy, the efficacy does not currently meet the commercial demands. In addition, the possible action modes of those biological agents have not been fully elucidated [6]. Therefore, a further exploration of the molecular mechanism underlying the interactions between P. italicum and citrus fruits is crucial for the development of novel strategies to control citrus blue mold.
During the long-term arms race between pathogens and hosts, plants have developed two sophisticated layers of immune systems for many pathogens: pathogen-associated molecular pattern (PAMP)-triggered immunity (PTI) and effector-triggered immunity (ETI) [7]. In order to escape plant immune system, pathogens are forced to secrete effectors for a successful colonization in the hosts [8]. These effectors can be secreted proteins or small molecules, most of which serve as virulence factors, PAMPs, toxin proteins,

Sample Preparation, RNA-Seq Library Construction, and Illumina Sequencing
Two wounds (1 × 2 mm) were created uniformly at the equator of the citrus fruit with a 0.5 mm diameter needle. A 10 µL conidia suspension of P. italicum (1 × 10 6 spores mL −1 ) was injected into each wound. The inoculated fruits were placed in transparent sealed boxes and stored at 26 • C. The tissue samples (about 3 g) were collected from the wound after 72 hpi. They were immediately frozen with liquid nitrogen and stored at −80 • C for subsequent analysis. Three replicates were prepared for the sample. RNA was extracted using a plant total RNA extraction kit (Aidlab Biotechnologies Co., Ltd., Beijing, China), according to the manufacturer's instructions (with three biological replicates). The Illumina HiSeq 2000 sequencing platform was used for sequencing total RNA, which was conducted by the BGI Company (Shenzhen, China). The raw RNA-Seq data files were submitted to the NCBI database under SRA accession number SRP362092.

Effector Protein Prediction Pipeline
Based on the results of secretome prediction and RNA-seq data, secreted proteins highly induced by infected Navel oranges were selected with an expression level greater than an average expression level (the average FPKM value of all genes detected in the genome of P. italicum) [24]. According to the method of Levin et al. [24], with slight modifications, the highly induced secreted proteins were classified according to the following criteria: (a) small (<400 aa) and cysteine-rich (>4) proteins; (b) proteins with gene function annotations or belonging to known protein families; (c) proteins predicted to be effectors using EffectorP-fungi (v3.0; http://effectorp.csiro.au/, accessed on 30 September 2020) software [21]; and (d) proteins belonging to known effector families or with effector domains that have been previously reported in the literature. Proteins fulfilling two or more of the above characteristics were selected as the optimal candidate effectors.

RNA Extraction and cDNA Synthesis
The tissue samples from the wounded sites of Navel oranges were collected at 24, 48, and 72 hpi using 0 hpi as a control. The mycelium samples from PDA medium at 26 • C were collected at 24, 48, 72, and 96 hpi, respectively. P. italicum spores that were frozen immediately after culturing on PDA medium at 26 • C for one week were prepared as a control.
The total RNA of P. italicum mycelia and Navel orange tissues were extracted using an RNA kit (Omega Bio-tek, Doraville, GA, USA) and plant total RNA extraction kit, respectively, according to the manufacturer's instructions. cDNA was synthesized with a HiScript ® II Q RT SuperMix for qPCR (+gDNA wiper) reverse transcription kit (Vazyme Biotechnology Co., Ltd., Nanjing, China), following the manufacturer's protocols. Each experiment was repeated three times.
RT-qPCR was performed with a CFX96 Real-Time PCR Detection System (Analytik Jena, Germany), according to the manufacturer's protocols. Each 25 µL of PCR reaction contained 5 µL of ChamQ SYBR qPCR Master Mix (Vazyme Biotechnology Co., Ltd., Nanjing, China), 1 ng-100 ng cDNA, and 0.2 µM of each primer. The reaction procedure was 95 • C 30 s, followed by 40 cycles of 5 s at 95 • C, 30 s at 60 • C; melt curve. Samples collected at 0 hpi were used to calibrate expression levels. The β-actin gene of P. italicum was used as an internal reference [17]. A combination of three biological replicates with three technical replicates of each reaction were used. The relative expression levels of genes were calculated by the 2 −∆∆Ct method [38].

Piwsc1 Sequence and Phylogenetic Analysis
The conserved domains of P. italicum PiWSCs were analyzed by InterPro (v5.1; http: //www.ebi.ac.uk/interpro/, accessed on 29 March 2022) [39]. To establish the evolutionary relationship of P. italicum Piwsc1, the amino acid sequence of P. italicum Piwsc1 was analyzed by BLASTp, and 20 species from 10 genera were selected; then the phylogenetic tree of P. italicum Piwsc1 was constructed with MEGA11 software using the Neighbor Joining method [40]. The bootstrap consensus tree was inferred from 1000 replicates.

Construction and Verification of Piwsc1 Knockout and Complementation Mutant
Gene disruption by homologous recombination was performed, as described earlier by Catlett et al. [41]. For Piwsc1 knockout, flanking regions of the 5 -region (1467 bp) and 3 -region (1281 bp) of the Piwsc1 gene (PITC_087410) were PCR amplified from the genomic DNA of wild-type P. italicum (WT) using wsc1up-F/R and wsc1do-F/R primer pairs, respectively (Table S2). Figure S1a presents the knockout cassette construction strategy. The knockout box was constructed by the fusion of the 5 -region, resistance gene HygB, and 3 -region, and connected to the pMD19-T vector to obtain the pMD-Piwsc1 plasmid. Firstly, the pMD-Piwsc1 vector and pCAMBIA3300 plasmid were digested with SmaI and HindIII, then ligated with T4 ligase and transformed into Escherichia coli DH5α to obtain the Piwsc1 gene knockout vector.
Piwsc1 complementation mutants were constructed using the protocol previously described by Yang et al. [11] with slight modifications. The full-length Piwsc1 gene was amplified from WT genomic DNA. The fragments of the 5 -region (wsc1up5-F/R, 1137 bp), resistance gene G418 (G418-F/R, 1221 bp), TrpC (TrpC-F/R, 368 bp), and Pi-wsc1+3 -region (wsc1-F/wsc1do3-R, 1984 bp) were fused to construct the complementation box ( Figure S1b), and then the complementation vector of the Piwsc1 gene was obtained according to the process of constructing a knockout vector.
The selected vectors were introduced into Agrobacterium tumefaciens AGL-1 cells, which were subsequently used to transform P. italicum, as previously described [42]. The obtained transformants were screened for resistance using the corresponding antibiotics (knockout transformants: 50 µg/mL hygromycin B/complementation transformants: 30 µg/mL geneticin). Select positive transformants to extract genomic DNA by the CTAB method [43], and use specific detection primers for PCR and RT-PCR identification (Table S2).

Phenotype of the Knockout Transformant (∆Piwsc1) and Complementation Transformant (∆Piwsc1-co)
Radial growth was examined following the protocol described by Yang et al. [29]. The fungus disc was made with a 5 mm punch and inoculated onto a PDA plate. The radial growth of the cultures was measured daily. Spore germination was evaluated by the calculation of percent germination using a ZEISS microscope (Zeiss Co., Oberkochen, Germany). Microscope images of the germinating spores were obtained following 10 h of incubation at 26 • C of 10 µL spore suspension (3 × 10 6 spores mL −1 ) inoculated on 0.5% agar-PDA plates. Sporulation was determined according to Levin et al. [24]. A total of 3 mL of sterile distilled water was added to three-days-old PDA plates; each plate was washed with an additional 2 mL of distilled water to ensure maximum spore collection. The resulting spore suspensions were diluted and counted with a hemocytometer. The number of spores per plate was calculated as spore concentration (spores mL −1 ) × dilution factor × 5 mL.

Virulence Assay
The virulence assay was performed according to Yang et al. [44]. The fruits were inoculated with P. italicum as described above. Five Navel oranges constituted a biological replicate and three replicates were used for each treatment. The average decay diameters of each treatment were monitored at 3, 4, 5, and 6 d post-inoculation (dpi).

Statistical Analysis
All statistical analyses were performed by one-way analysis of variance (ANOVA) using SPSS 22.0 statistical software (SPSS Inc., Chicago, IL, USA). Mean separations were performed by Fisher's least significant difference (LSD) test (p < 0.05). Different letters indicated significant difference. Bars indicated standard error.

Prediction of P. italicum Secretome
The genome sequence and annotation of the P. italicum strain PHI-1 was previously described by Ballester et al. [30]. The genome contains a total of 9996 protein sequences, which were used to predict and identify secreted proteins. SignalP v5.0, TargetP v1.1, TMHMM v2.0, and PredGPI were collectively employed to predict the total secretome ( Figure 1a, Table S3). Six hundred and fifty-four proteins were identified with signal peptides by analyzing the total protein sequences through SignalP v5.0. After excluding the proteins containing either a chloroplast transit peptide or a mitochondrial targeting peptide by TargetP v1.1, 531 proteins were obtained with extracellular localization signals, of which 477 proteins were selected by TMHMM v2.0 analysis that had no or only one harbor transmembrane helix. Using PredGPI to predict the proteins that harbor glycophosphatidylinositol anchor motifs, 102 proteins were excluded as surface proteins rather than secreted effectors. Eventually, a total of 375 secreted proteins were selected for further analysis, representing 3.8% of the P. italicum proteome. , and Venn analysis (c). In (c), 29 optimal candidate effector proteins, with five out of them being core candidate effector proteins, are identified. Green: proteins smaller than 400 amino acids with 4 or more cysteines; blue: proteins with gene function annotation or belonging to known protein families; pink: proteins predicted to be effectors by EffectorP-fungi; yellow: proteins belonging to known effector families or with effector domains that have been previously reported in the literature.

CAZymes of P. italicum Secretome
CAZymes play important roles in the pathogenicity of pathogenic fungi [45]. Among the 375 secreted proteins, 109 secreted proteins (29.07%) belonging to the CAZymes family were classified in 58 CAZymes subgroups by using the dbCAN database. Most of them are glycoside hydrolases (GHs), accounting for 57.8% ( Figure 2, Table S4). In the current study, seven carbohydrate-binding proteins that have been annotated as concanavalin Alike lectins (GH7, GH11, GH16, GH54) were identified as secreted proteins during the invasion of Sclerotinia sclerotiorum on rape [46]. In addition, three secreted proteins that are associated with cell-wall-degrading enzymes, including pectinesterase (CE8) and fungal chitosanase (GH75), are also included in the category of CAZymes. , and Venn analysis (c). In (c), 29 optimal candidate effector proteins, with five out of them being core candidate effector proteins, are identified. Green: proteins smaller than 400 amino acids with 4 or more cysteines; blue: proteins with gene function annotation or belonging to known protein families; pink: proteins predicted to be effectors by EffectorP-fungi; yellow: proteins belonging to known effector families or with effector domains that have been previously reported in the literature.

CAZymes of P. italicum Secretome
CAZymes play important roles in the pathogenicity of pathogenic fungi [45]. Among the 375 secreted proteins, 109 secreted proteins (29.07%) belonging to the CAZymes family were classified in 58 CAZymes subgroups by using the dbCAN database. Most of them are glycoside hydrolases (GHs), accounting for 57.8% ( Figure 2, Table S4). In the current study, seven carbohydrate-binding proteins that have been annotated as concanavalin A-like lectins (GH7, GH11, GH16, GH54) were identified as secreted proteins during the invasion of Sclerotinia sclerotiorum on rape [46]. In addition, three secreted proteins that are associated with cell-wall-degrading enzymes, including pectinesterase (CE8) and fungal chitosanase (GH75), are also included in the category of CAZymes.

Analysis of P. italicum Candidate Effector Proteins
We used the commercial platform to establish the RNA-seq library. RNA-seq experiments were conducted using interface tissues from wild-type P. italicum-infected Navel oranges. The number of expressed genes detected on the Illumina HiSeq sequencing plat-

Analysis of P. italicum Candidate Effector Proteins
We used the commercial platform to establish the RNA-seq library. RNA-seq experiments were conducted using interface tissues from wild-type P. italicum-infected Navel oranges. The number of expressed genes detected on the Illumina HiSeq sequencing platform was 8766. The FPKM values of 375 secreted proteins in RNA-Seq are listed in Table S5.
Based on the predicted secreted proteins of P. italicum, we screened the genes that were highly expressed on infected Navel oranges, resulting in 68 genes in our prediction pipeline (Figure 1b). The highly expressed secreted proteins were classified as follows.

Small and Cysteine-Rich Highly Expressed Secreted Proteins
When screening highly expressed secreted proteins by length and the number of cysteine residues, 27 secreted proteins matched the conditions of small-sized (<400 aa) and cysteine-rich (>4) ( Table S6). Half of these proteins were 100-200 amino acid residues in length and most of the effector proteins had 4 and 6 cysteine residues.

Highly Expressed Secreted Proteins by EffectorP-fungi Software Prediction
Through EffectorP-fungi software, 24 of 68 secreted proteins were found to meet the EffectorP effector protein criteria (Table S8); 15 were predicted as apoplastic effectors, 4 were predicted as cytoplasmic effectors, and 5 were predicted either as apoplastic or cytoplasmic effectors.
As mentioned above, proteins meeting two or more of the above criteria were selected to be analyzed further as optimal candidates (Figure 1c), which resulted in 29 candidate effectors (Table 1). Among them, five annotated proteins belonged to the glycoside hydrolase family (PITC_048860, PITC_077450, PITC_005000, PITC_081180, and PITC_020870). Another five proteins were annotated as peptidase (PITC_061260, PITC_079910, PITC_007270, PITC_014210, and PITC_008610), in which one (PITC_061260) had aspartic-type endopeptidase activity and the other four had a serine-type carboxypeptidase activity. PITC_043760 and PITC_047900 presented RNase activity. Two hydrophobic proteins (PITC_001010 and PITC_015600) were considered to be the structural constituents of the cell wall, and one (PITC_034160) was a pectate lyase. PITC_013620 belonged to the p24 (emp24/gp25L) family of transmembrane proteins. PITC_019680 was annotated as thioredoxin. PITC_081470 belonged to phosphatase. PITC_099400 contained a myeloid differentiation factor 2-related lipid-recognition (ML) domain that was involved in innate immunity process and sterol transport. PITC_097880 encoded a NIP and possessed a nucleotide pyrophosphatase/ phosphodiesterase1 (NPP1) domain. PITC_087410 contained a WSC domain that participated in the cell wall integrity (CWI) signaling pathway. PITC_016950 belonged to ceratoplatanin proteins. In addition, seven were hypothetical proteins with no functional annotation, but highly expressed during infection in citrus fruits (PITC_045800, PITC_051450, PITC_068580, PITC_014450, PITC_085860, PITC_014290, and PITC_062780). Five of them, namely, carbohydrate-binding WSC (PITC_087410), cerato-platanin (PITC_016950), pectin lyase (PITC_034160), ribonuclease_T2 (PITC_043760), and guanine-specific ribonuclease N1/T1 (PITC_047900), met all of the above criteria, and thus were regarded as core effector proteins (Table 1).

RT-PCR and RT-qPCR Analyses of Candidate Effector Genes
To validate whether these effectors of P. italicum were related to the early stage of infection, the time lines of blue mold symptoms and disease development in Navel oranges were monitored at 24, 48, and 72 hpi (Figure 3). At 24 hpi, the wound sites of the fruits began to collapse and the wound edges were in a water-soaked soft-rot state. A small amount of mycelium on the soft-rot sites appeared at 48 hpi. The soft rot rapidly expanded by 72 hpi and conidia generated on the mycelia ( Figure S2). Symptomatically, time points between 48 hpi and 72 hpi seemed to be critical for P. italicum infection.
The time course of expression patterns of twenty-nine candidate genes were studied using RT-PCR with total RNAs extracted from P. italicum-infected Navel oranges at 0, 24, 48, and 72 hpi ( Figure S3). A total of 24 effector genes were detected at 24 hpi, indicating that they might be involved in the early process of P. italicum for invasion and colonization on citrus fruits. The remaining 5 candidate effector genes (PITC_051450, PITC_085860, PITC_001010, PITC_015600, and PITC_097880) began to express at 48 hpi, which might be required for a late pathogenicity.
of effector genes in vitro were not always consistent with those in vivo. For example, the expression of PITC_062780 reached its peak at 24 hpi in vitro, whereas its expression was not detected in vivo at this time. The expression levels of the three selected core effector genes PITC_047900, PITC_087410, and PITC_016950 reached peaks at 24 hpi, 48 hpi, and 48 hpi in vitro, respectively, while their expression patterns exhibited an increasing tendency in vivo. Taken together, these results suggest that these core effectors might play important roles in the pathogenicity of this pathogen. Based on the above analyses, PITC_087410 (denoted as PiWSC1) matched all the criteria for candidate effector proteins and were higher expressed in vitro (24-96 hpi) and in vivo (24-72 hpi) conditions, which was then selected for further functional analysis.

Characterization of PiWSC1
The Piwsc1 gene encodes a protein of 303 aa residues with the full-length of 976 bp, which contains one intron. Except for PiWSC1, PITC_001070 (denoted as PiWSC2) and PITC_052160 (denoted as PiWSC3) were also annotated as carbohydrate-binding WSC proteins in the genome of P. italicum (Figure 4a). PiWSC2 is a GPI-anchored protein, and PiWSC3 shows a reliability score of more than 2 by TargetP analysis; both of them were excluded as secreted proteins. The structural domains analysis by InterPro showed that all three WSC proteins had a signal peptide at the N-terminal, and PiWSC1 and PiWSC3 Twelve randomly selected candidate effector genes were further investigated by RT-qPCR for their expression on PDA culture (in vitro) (Figure 3a) and during Navel orange infection (in vivo) (Figure 3b). We found that all 12 candidate effector genes were actively expressed both in vitro and in vivo. On the PDA culture, 9 effector genes exhibited the highest expression levels at 48 hpi, with 2 effector genes (PITC_062780 and PITC_047900) reaching peaks at 24 hpi and 1 effector gene (PITC_097880) at 72 hpi. During Navel orange infection, the expression patterns of 8 effector genes demonstrated an upward tendency, with 3 effector genes (PITC_068580, PITC_013620, and PITC_081180) reaching their peaks at 24 hpi, and 1 effector gene (PITC_019680) at 48 hpi. However, the expression patterns of effector genes in vitro were not always consistent with those in vivo. For example, the expression of PITC_062780 reached its peak at 24 hpi in vitro, whereas its expression was not detected in vivo at this time. The expression levels of the three selected core effector genes PITC_047900, PITC_087410, and PITC_016950 reached peaks at 24 hpi, 48 hpi, and 48 hpi in vitro, respectively, while their expression patterns exhibited an increasing tendency in vivo. Taken together, these results suggest that these core effectors might play important roles in the pathogenicity of this pathogen.
Based on the above analyses, PITC_087410 (denoted as Piwsc1) matched all the criteria for candidate effector proteins and were higher expressed in vitro (24-96 hpi) and in vivo (24-72 hpi) conditions, which was then selected for further functional analysis.

Characterization of Piwsc1
The Piwsc1 gene encodes a protein of 303 aa residues with the full-length of 976 bp, which contains one intron. Except for Piwsc1, PITC_001070 (denoted as PiWSC2) and PITC_052160 (denoted as PiWSC3) were also annotated as carbohydrate-binding WSC proteins in the genome of P. italicum (Figure 4a). PiWSC2 is a GPI-anchored protein, and PiWSC3 shows a reliability score of more than 2 by TargetP analysis; both of them were excluded as secreted proteins. The structural domains analysis by InterPro showed that all three WSC proteins had a signal peptide at the N-terminal, and Piwsc1 and PiWSC3 had a cytoplasmic domain at the C-terminal. In addition, Piwsc1 had an SKG6 domain, and PiWSC3 had a transmembrane domain of the epidermal growth factor receptor family of protein tyrosine kinases (TM_EGFR). By contrast, PiWSC2 did not appear to possess any other conserved domains. The protein sequence alignment results show that the three proteins have a low sequence identity between them ( Figure S4). of protein tyrosine kinases (TM_EGFR). By contrast, PiWSC2 did not appear to possess any other conserved domains. The protein sequence alignment results show that the three proteins have a low sequence identity between them ( Figure S4). PiWSC1 is conserved in genus Penicillium and other fungi genera. The phylogenetic tree showed that P. italicum PiWSC1 had the highest homology with P. expansum and a higher sequence identity with the homologues of the Aspergillus species than other filamentous fungi (Figure 4b).

Functional Analysis of PiWSC1
To further explore the biological function of Piwsc1 in P. italicum, Piwsc1 knockout and complementation mutants of P. italicum were constructed. The mutants were verified by PCR and RT-PCR using specific primers (Figures S1c-e). A knockout mutant, ΔPiwsc1, and a complementation mutant, ΔPiwsc1-co, were then selected for functional analysis.
Compared to the WT, the radial growth rate of ΔPiwsc1 was decreased by 20.15% at 7 dpi (Figure 5a). The colony morphology of ΔPiwsc1 was also altered, as manifested by the decreased hyphal edge density (Figure 5b). The spore germination rate was reduced by 22.27% at 10 hpi (Figure 5c), and the sporulation was reduced by 35.31% (Figure 5d). These findings indicate that Piwsc1 plays an important role in the growth and development of P. italicum. Piwsc1 is conserved in genus Penicillium and other fungi genera. The phylogenetic tree showed that P. italicum Piwsc1 had the highest homology with P. expansum and a higher sequence identity with the homologues of the Aspergillus species than other filamentous fungi (Figure 4b).

Functional Analysis of Piwsc1
To further explore the biological function of Piwsc1 in P. italicum, Piwsc1 knockout and complementation mutants of P. italicum were constructed. The mutants were verified by PCR and RT-PCR using specific primers ( Figure S1c-e). A knockout mutant, ∆Piwsc1, and a complementation mutant, ∆Piwsc1-co, were then selected for functional analysis.
Compared to the WT, the radial growth rate of ∆Piwsc1 was decreased by 20.15% at 7 dpi (Figure 5a). The colony morphology of ∆Piwsc1 was also altered, as manifested by the decreased hyphal edge density (Figure 5b). The spore germination rate was reduced by 22.27% at 10 hpi (Figure 5c), and the sporulation was reduced by 35.31% (Figure 5d). These findings indicate that Piwsc1 plays an important role in the growth and development of P. italicum.
To determine the role of the Piwsc1 gene in pathogenicity, Navel oranges were inoculated with WT, ∆Piwsc1, and ∆Piwsc1-co strains, respectively, and the symptoms of fruit decay were recorded. At the initial stage of infection with ∆Piwsc1, the spore formation and the degree of soft rot near the infection site were much lower than WT and ∆Piwsc1-co (Figure 5e), and the diameter of citrus lesions inoculated with ∆Piwsc1 decreased by 57.14%, compared to WT and ∆Piwsc1-co at 6 dpi (Figure 5f). The results demonstrate that Piwsc1 plays an important role in the virulence of P. italicum. Figure 5. Effect of Piwsc1 deletion on P. italicum radial growth, sporulation, spore germination, and virulence. Radial growth diameter changes (a) and colony morphology at 7 dpi (b), spore germination rate at 10 hpi (c), and sporulation at 3 dpi (d) of WT, ΔPiwsc1, and ΔPiwsc1-co strains on PDA plates are recorded. Rot status at 3 and 6 dpi (e) and lesion diameters from 3 to 6 dpi (f) of WT, ΔPiwsc1, and ΔPiwsc1-co strains on Navel oranges are recorded.
To determine the role of the Piwsc1 gene in pathogenicity, Navel oranges were inoculated with WT, ΔPiwsc1, and ΔPiwsc1-co strains, respectively, and the symptoms of fruit decay were recorded. At the initial stage of infection with ΔPiwsc1, the spore formation and the degree of soft rot near the infection site were much lower than WT and ΔPiwsc1co (Figure 5e), and the diameter of citrus lesions inoculated with ΔPiwsc1 decreased by 57.14%, compared to WT and ΔPiwsc1-co at 6 dpi (Figure 5f). The results demonstrate that Piwsc1 plays an important role in the virulence of P. italicum.

Discussion
Effector proteins are important weapons of pathogenic fungi to compete with plants and cause infection [19,55]. The interaction mechanism between P. italicum and citrus fruits at the molecular level was studied to predict and confirm pathogenic effector proteins through bioinformatic analysis, transcriptomic patterns during the infection, and the in vitro growth of deletion mutants from one core gene. A total of 375 secreted proteins of P. italicum were predicted by SignalP, TargetP, TMHMM, PredGPI, and other software. Of the screened 29 candidate effector proteins, 5 proteins were finally considered as core effectors that met 4 requirements of the criteria designed in the pipeline. We predicted that secreted proteins accounted for 3.8% of the total proteins in P. italicum, which was Figure 5. Effect of Piwsc1 deletion on P. italicum radial growth, sporulation, spore germination, and virulence. Radial growth diameter changes (a) and colony morphology at 7 dpi (b), spore germination rate at 10 hpi (c), and sporulation at 3 dpi (d) of WT, ∆Piwsc1, and ∆Piwsc1-co strains on PDA plates are recorded. Rot status at 3 and 6 dpi (e) and lesion diameters from 3 to 6 dpi (f) of WT, ∆Piwsc1, and ∆Piwsc1-co strains on Navel oranges are recorded.

Discussion
Effector proteins are important weapons of pathogenic fungi to compete with plants and cause infection [19,55]. The interaction mechanism between P. italicum and citrus fruits at the molecular level was studied to predict and confirm pathogenic effector proteins through bioinformatic analysis, transcriptomic patterns during the infection, and the in vitro growth of deletion mutants from one core gene. A total of 375 secreted proteins of P. italicum were predicted by SignalP, TargetP, TMHMM, PredGPI, and other software. Of the screened 29 candidate effector proteins, 5 proteins were finally considered as core effectors that met 4 requirements of the criteria designed in the pipeline. We predicted that secreted proteins accounted for 3.8% of the total proteins in P. italicum, which was consistent with reported 3.0%, 4.1%, 4.7%, and 5.1% for P. infestans, Penicillium digitatum_PHI26, P. italicum GL_Gan1, and S. sclerotiorum, respectively [17,46,56]. A large portion of secreted proteins in P. italicum are enzymes that function as glycosyl hydrolases, proteinases, peptidases, RNases, and pectin lyases (GH55, PL1, and PL3). Pectin lyases were regarded as virulence factors in a number of species due to their cell-wall-degrading activities [57]. These results are consistent with the previous studies predicting that Fusarium graminearum secretes various putative enzymes that degrade different components of host cells [58].
Some studies have confirmed that secreted proteins expressed during the infection stage are more likely to be pathogenic effectors [59]. For example, the expression levels of NEP1like Tal6 and LysM effectors were highly expressed during the early stages of infection caused by various fungi [10,27,60,61]. Notably, some effector genes of plant pathogens seem to be solely expressed during host colonization [62,63], and others, such as PePRT and PeLysMs effectors in P. expansum, can be expressed during apple colonization and in vitro growth [24,64]. In this study, while the randomly selected 12 effector genes were actively expressed in vitro and in vivo, we also noted that the induced expression levels in vivo were higher than those in vitro (Figure 3).
Among the 29 effector proteins of P. italicum, 18 conserved domains [46] were identified, such as NPP1, WSC, RNase, and cerato-platanin. Protein PITC_097880 was annotated as NIP that contained an NPP1 domain. The most thoroughly studied NIPs were within a family of non-catalytic NIPs. They were collectively named as necrosisand ethylene-inducing peptide 1-like proteins (NLPs) [16]. NLPs were first identified as the elicitors of cell death in dicotyledonous plants, which might promote plant infection by necrotrophic pathogens and stimulate plant innate immunity [65]. PITC_047900 and PITC_043760 were annotated as RNases, which might be related to pathogenicity since a homology protein-Fg12 from F. graminearum contributed to pathogen virulence and induced plant cell death [52]. PITC_016950 belongs to cerato-platanin proteins, a phytotoxic protein secreted by filamentous fungi. In S. sclerotiorum, SsCP1 is an important virulence factor and is recognized by plants to trigger plant defense responses [11]. Functional annotations were difficult for 7 effector proteins in this study since no conserved domain was found, although they were highly expressed during the P. italicum-citrus interaction. These group of proteins were predicted as apoplastic or cytoplasmic effector proteins on the EffectorP-fungi platform. Chen et al. [66] used the A. tumefaciens-mediated Nicotiana benthamiana transient expression system to verify the functions of 3 such effectors in Fusarium sacchari. The results showed that Fs00367 and Fs00597 suppressed BAX-induced cell death, while Fs05897 induced cell death. This indicated that the hypothetical proteins might also function as effector proteins.
Generally speaking, the low molecular weight and richness of cysteines are two important characteristics of effector proteins, but 7 effector proteins did not accord with these features in our case. For example, the PITC_014210, PITC_008610, PITC_079910, and PITC_007270 encoded serine proteases with a length over 500 amino acids. Many reports have shown that serine proteases play a key role in the pathogenicity of plant pathogens. The first form of evidence that subtilisin (a very diverse family of serine peptidases) is involved in the plant-pathogen interaction was found in tomatoes by Granell et al. [67], who found that subtilisin P69 accumulated in tomato leaves after the tomatoes were treated by citrus exocortis viroid. Furthermore, the Plasmodiophora brassicae serine protease Pro1 was identified as a member of the S28 protease family and was previously characterized to play an important role in stimulating dormant spore germination [54]. PITC_020870 was annotated as concanavalin A-like lectin/glucanase with only two cysteines, but it was predicted to be an apoplastic effector (Table S8). Similarly, the AvrLm1 effector of L. maculans was localized in the apoplast with only one cysteine [26]. This indicates that the high molecular weight and containing a small number of cysteines may also be an effector protein.
A core candidate effector, Piwsc1, was selected to verify its role in P. italicum infection in this study. WSC proteins were regarded as candidate effector proteins in P. expansum colonization on apples and identified as secreted proteins of entomopathogenic fungus Metarhizium anisopliae when grown under submerged fermentation in the presence of chrysalis as an inducer [24,47]. Similar to homologs in other fungi, the RT-qPCR results exhibited that Piwsc1 was highly induced during the P. italicum infection of citrus fruits. The deletion of Piwsc1 in P. italicum affected in vitro growth and infection. The growth rate, spore germination rate, and sporulation of ∆Piwsc1 were lower than those of WT and ∆Piwsc1-co, which are similar to other fungi, such as Aspergillus nidulans, in which the growth and sporulation were inhibited after the deletion of homologous protein genes wscA and wscB [68]. Moreover, the phenotypic changes in ∆Piwsc1 were mainly characterized by lower mycelial edge density compared to the WT and ∆Piwsc1-co strains. Similar results were also shown in Aspergillus fumigatus [69]. WSC proteins were described as sensors that were involved in the CWI pathway and stress responses [70,71]. A recent study revealed that proteins containing WSC domains have carbohydrate-binding capacities. In rice blast pathogen Pyricularia oryzae, the WSC domain in an alcohol oxidase PoAlcOX directed the attachment to xylan and fungal chitin/β-1,3-glucan [72]. Another example is root endophyte Serendipita indica SiWSC3, a lectin-like member transcriptionally induced in planta, binds to long-chain β-1-3-glucan, efficiently agglutinates fungal cells, and is additionally induced during fungus-fungus confrontation [73]. More evidence suggests that WSCs are involved in the processes of host-fungal interactions and fungus-fungus confrontation. Tong et al. [74] found that the deletion of wsc1I reduced the conidial infectivity and virulence of entomopathogenic fungi Beauveria bassiana to Galleria mellonella larvae. Our results for ∆Piwsc1-infected Navel oranges also exhibit a significant reduction in virulence, suggesting its active roles in P. italicum pathogenesis; however, the underlying mechanism needs further investigation.

Conclusions
In the current study, the potential effector proteins in P. italicum, a common fungal species causing postharvest blue mold of citrus fruits, were characterized. We designed a prediction pipeline to predict the effector proteins of P. italicum and used this pipeline to screen out 29 optimal candidate effectors. Our results show that 12 randomly selected candidate effector genes are highly expressed during the early stages of growth on PDA and infection on citrus fruits. Functional analysis was performed on a core candidate effector protein, Piwsc1, and the results demonstrate that the deletion of Piwsc1 decreases the virulence of this fungus on citrus fruits. Our study on the effector proteins of P. italicum provides a theoretical basis for further exploring the molecular mechanism of the citrus-P. italicum interaction.